Based on the FNL reanalysis data and U.S. Joint Typhoon Warning Center(JTWC) tropical cyclone data,the WRF(Weather Research and Forecasting) model is employed to establish the ensemble prediction system for the typhoon Saola(1209) based on the initial perturbation methods of Breeding of Growing Mode(BGM) and Ensemble Transform Kalman Filter(ETKF),respectively.The ensemble forecasting experiments for the typhoon track are conducted.It is detected that the ensemble mean forecasting results using either BGM or ETKF perform much better than control forecasting results in the aspects of wind,temperature and geopotential height fields.And the ETKF method is superior to the BGM method,especially in the aspects of wind and temperature fields.Results show that the dispersions of ensemble forecast system using BGM method are smaller than those using ETKF method.The BGM ensemble forecast system has a fairly limited ability to describe the whole atmosphere.Additionally,the ensemble mean track forecasts based on BGM and ETKF are both superior to the control forecast,and well manage the problem of faster speed after the second landfall in Fujian coast for the control forecast.However,for the landfall location,the predictions by means of the two methods are not so satisfactory.In addition,the ETKF method is considerably better than the BGM method in the ensemble track prediction of typhoon Saola.